Article Text

Extended report
A vasculitis centre based management strategy leads to improved outcome in eosinophilic granulomatosis and polyangiitis (Churg–Strauss, EGPA): monocentric experiences in 150 patients
  1. Frank Moosig1,
  2. Jan Phillip Bremer1,
  3. Bernhard Hellmich2,
  4. Julia Ulrike Holle1,
  5. Konstanze Holl-Ulrich3,
  6. Martin Laudien4,
  7. Christine Matthis5,
  8. Claudia Metzler6,
  9. Bernhard Nölle7,
  10. Gert Richardt8,
  11. Wolfgang L Gross1
  1. 1Department of Rheumatology, University Hospital Schleswig-Holstein and Klinikum Bad Bramstedt, Bad Bramstedt, Germany
  2. 2Department of Internal Medicine, University of Tübingen and Hospital Plochingen, Plochingen, Germany
  3. 3Department of Pathology, University of Lübeck, Lübeck, Germany
  4. 4Department of Otorhinolaryngology, Head and Neck Surgery, University of Kiel, Kiel, Germany
  5. 5Department of Social Medicine, University of Lübeck, Lübeck, Germany
  6. 6Department of Internal Medicine, Hospital Barmherzige Brüder, Regensburg, Germany
  7. 7Department of Ophthalmology, University of Kiel, Kiel, Germany
  8. 8Department of Cardiology, Hospital Bad Segeberg, Bad Segeberg, Germany
  1. Correspondence to Dr Frank Moosig, Department of Rheumatology, University Hospital Schleswig-Holstein and Klinikum Bad Bramstedt, Oskar-Alexander-Street 26, Bad Bramstedt 24576, Germany, f.moosig{at}


Objective To evaluate a vasculitis centre based management strategy for eosinophilic granulomatosis and polyangiitis (Churg-Strauss, EGPA).

Methods A retrospective cohort study at a vasculitis referral centre was performed. All EGPA patients admitted from 1990 to 2009 were included. A structured interdisciplinary work-up for proof of diagnosis, Disease Extent Index and Birmingham Vasculitis Activity Score was performed. Immunosuppressive therapy was initiated and regularly adapted. Treatment targets were induction and maintenance of remission according to definitions given by the European League Against Rheumatism and the European Vasculitis Study Group. Outcomes were mortality, rate of remission, relapses, adverse events and prednisolone-dose.

Results Out of 269 patients with suspected EGPA 150 fulfilled the inclusion criteria. Of those, 104 had more than one follow-up visit resulting in a mean follow up of 53±4.9 months. By using additional data sources the follow-up concerning survival was extended to 92±5 month. Severe organ manifestations occurred at heart (46%), kidney (18%) and lungs (10%). Cyclophosphamide was used in 107 patients (71%). The prednisolone-doses of all patients were within the targeted range (i.e. ≤7.5mg) in 69% of the total follow-up time; the median dose at end of follow-up was 5mg/d. The 10-year survival rate was 89% resulting in mortality comparable to the general population (SMR 1.29). Only patients with cardiac failure associated with EGPA had an increased mortality (SMR 3.06).

Conclusions Regular re-evaluation and target-orientated adaption of therapy may lead to normalization of life expectancy and attenuation of disease progression. Continued centre based interdisciplinary treatment should be standard of care.

  • Systemic vasculitis
  • Outcomes research
  • Multidisciplinary team-care

Statistics from


Eosinophilic granulomatosis and polyangiitis (EGPA) is the rarest disease out of the group of ANCA-associated vasculitides.1 Severe courses are frequent and require intense immunosuppression. Many patients need high glucocorticoid (GC) doses over long periods of time with an associated high burden of adverse events.2 Retrospective studies report a 10-year mortality of up to 30%.3 As data from controlled trials in EGPA are sparse, diagnostic and therapeutic principles are often adapted from other ANCA-associated vasculitides. However, until now there is no proof that intense immunosuppression is adequate in non-life threatening EGPA or for the purpose of GC-sparing and improvement of outcome. As the rarity of EGPA hinders the systematic and controlled evaluation of single interventions in prospective and controlled clinical trials it seems adequate to describe whole complex management strategies. The patients at our centre have been treated according to such a strategy, including consequent GC-sparing. We herein retrospectively analysed the outcomes of this cohort. Although, it is not possible to determine the contribution of single aspects of this complex intervention the whole approach was found to be successful.

Patients and methods

Patients attending our centre from January 1990 to December 2009 were included in this retrospective analysis if the following requirements were fulfilled: histological and/or clinical evidence of vasculitis and fulfilment of the American College of Rheumatology classification criteria for EGPA4 as well as exclusion of other underlying causative conditions.

All patients underwent an initial interdisciplinary work-up and follow-up visits at a regular basis as described earlier (5; see online supplementary table S1). To complete data concerning mortality for patients without follow-up visit within the last 6 months, additional evidence, for example, reports from co-treating physicians were used.

Ethics approval

The reported approach represents the inhouse standard management. The patients gave informed written consent for the use of their data. The data were analysed pseudonymously.

Outcome measures

The results of the work-ups and changes in medication were documented. Since 1994, the Birmingham Vasculitis Activity Score (BVAS) was calculated. For visits before 1994, it was recalculated retrospectively.6 Additionally, the Disease Extent Index (DEI) was used.7 The definitions for disease stages, remission, relapse, response and refractory disease were used prospectively as published elsewhere (8; see online supplementary table S2) and are largely identical with those suggested by the European Vasculitis Study Group (EUVAS)/EULAR.9 Since 1996, the Five Factor Score (FFS) was used.10 The vasculitis damage index was calculated retrospectively.


Medication was chosen according to activity and extent of disease and was regularly adapted (see online supplementary figure S1).

In non-organ- or life-threatening disease prednisolone monotherapy was used with an initial dose between 15 and 75 mg/day. Steroids were then tapered according to a fixed protocol. In case tapering to a dose of 7.5 mg/day within 3 months was not achieved, steroid-sparing medication was added (azathioprine (AZA), methotrexate (MTX), leflunomide (LF)). The choice of drug was based on previous and concomitant medication, comorbidities and organ function (eg, renal insufficiency). MTX dosage was between 10 and 30 mg (maximum 0.3 mg/kg) subcutaneous or intravenous weekly with folic acid substitution. LF was dosed 10–30 mg/day and AZA 2 mg/day kg body weight (BW).

In cases of life- or organ-threatening manifestations, defined as alveolar haemorrhage, severe eosinophilic alveolitis, that is, leading to respiratory (partial-) insufficiency and/or not responsive to GC, involvement of kidneys, heart, peripheral nervous system (PNS; only mononeuritis multiplex or sensomotoric neuropathy), central nervous system (CNS) or gastrointestinal tract, cyclophosphamide (CY) therapy and prednisolone (initially 1 mg/kg) were initiated as described earlier.2 ,11 CY therapy was either administered as oral (2 mg/day kg BW, usually for 3 months) or as pulse (15 mg/kg BW, usually six times) depending on the date of inclusion (before 2006 oral CY was preferred). Mesna and sulphametoxazol-trimethoprim prophylaxis were routinely used. Prednisolone was reduced early, starting 3–8 days after initiation. The aim was to reach a dose of below 15 mg/day after 2–3 months and below 7.5 mg/day when switching to maintenance medication.

In refractory EGPA, different salvage therapies in combination with GC were applied as described earlier.12–14

If remission was attained, AZA, MTX or LF in the above mentioned doses was initiated. A prednisolone dose ≤7.5 mg/day was part of the definition for remission. During maintenance therapy, prednisolone dose was further tapered slowly with a target dose of 2–3 mg/day. In cases of long-term remission, defined as at least 1 year without flare, maintenance medication in a few cases was tapered or stopped on the basis of an individual decision by the treating physician and the patient.

Disease relapses were treated as described above according to severity. Patients were regularly re-evaluated and therapy was adapted if the target was not reached.

Prednisolone doses and dose changes were documented at each visit allowing for an estimation of cumulative doses and median daily use. Furthermore, patients were screened for prednisolone-related side effects at every visit including bone density measurement using Dexa method at baseline and then every 2 years in most patients.

All patients underwent at least one structured patient education programme15 with focus on how to recognise signs and symptoms of disease flares and of drug side effects. Furthermore, knowledge on necessary lab testing was provided.

Statistical analysis

Data are presented as means±SEM or median and ranges. Statistical analysis was performed using SPSS V.12.0 for windows (SPSS Inc., Chicago, Illinois, USA). The Mann–Whitney U test and the Fisher's exact test were used as appropriate. Mean survival, time to first relapse, time to remission and impact of additional parameters were analysed using the Kaplan–Meier life-table analysis and log-rank test, with 95% CI. For multivariate analysis Cox-regression was performed. Standardised mortality ratios were calculated as the ratio of observed to expected deaths. The expected numbers of deaths were obtained using age- and sex-specific death rates published in the mortality table of Germany 2006/2008 (Statistisches Bundesamt, Wiesbaden, Germany).


Demographic data

Of 269 admitted patients classified as EGPA or hypereosinophilic syndrome by the referring physician, 150 (74 female subjects) fulfilled the inclusion criteria. The mean age at diagnosis was 49.1±1.2 years. In 80, histological proof of vasculitis was available. The excluded patients were diagnosed with possible EGPA (n=62) without fulfilling the inclusion criteria, with hypereosinophilic syndrome (n=51) or with paraneoplastic conditions (n=5). One patient had parasitic infection.

In 132/150, immunosuppression was started prior to referral (81 prednisolone monotherapy, 16 oral and seven pulse CY, 15 AZA, 13 MTX, one interferon-α).

Baseline characteristics are given in table 1.

Table 1

Baseline characteristics of the EGPA cohort

Initial presentation

At initial presentation, 75/150 (50%) suffered from active organ- or life-threatening disease. PNS involvement was seen in 52 (35%), with mononeuritis multiplex being the dominant finding (35; 23%). Symptomatic lung involvement except asthma was seen in 27 (18%). In four of those cases (2.6%), alveolar haemorrhage was detected on broncho-alveolar-lavage. Active cardiac involvement in 26 (17%) consisted of one or two of the following: cardiomyopathy not attributable to other conditions (13), pericardial effusion (10) and prednisolone-sensitive arrhythmias (9). In all, 12 (8%) had clinically active gastrointestinal manifestations (pain, diarrhoea), with eosinophilic tissue infiltrations in all 10 available biopsies and additional proof of vasculitis in four. Glomerulonephritis as suggested by proteinuria >1 g/day and erythrocyte casts was diagnosed in 8 (5%). Detailed characteristics are given in table 1.

Organ involvement at baseline occurred in a high frequency (figure 1). Comorbidities are given in online supplementary table S3.

Figure 1

The figure shows the cumulative organ involvement at any time during the disease course as documented at onset by the initially treating hospital/physician (first bar), according to the result of the diagnostic work-up at the vasculitis centre (second bar) at referral and according to the results of the last available visit (third bar in each column). A, joints; B, constitutional symptoms; C, central nervous system; E, ENT tract; Ey, Eye; Gi, gastrointestinal tract; H, heart; K, kidney; L, lung; P, peripheral nervous system; S, skin.

Active disease without organ- or life-threatening manifestations was found in 40/150 (27%). Ear–nose–throat (ENT) involvement (n=23; 15%), musculoskeletal-(n=11; 7%) and constitutional symptoms (n=12; 8%) dominated.

At initial presentation, 35/150 (23%) were found to be in remission. Those patients mainly were referred to our centre for confirmation of diagnosis, to optimise drug therapy and to let patients attend the education programme.

Among the 45 (30%) Anti-Neutrophil-Cytoplasmic-Antibodies (ANCA)-positive patients, 38 had a perinuclear and seven a cytoplasmic pattern. On ELISA, 34 were positive for MPO- and three for PR3-ANCA. For a detailed comparison between ANCA-positive and ANCA-negative patients, see online supplementary table S4.


At the end of the observation period, 104/150 had at least one follow-up visit (mean 5.2, median 3, range 1–28), resulting in a mean follow-up of 53±5 months, that is, a total of 460 patient years. The median interval between visits was 6 months. Recruitment increased steadily over time (35 patients in 1990–1999 vs 115 patients in 2000–2009).

During that time, 70/104 (67.3%) attained remission. Overall, 24 (23%) reached a state of absent disease activity (BVAS=0) but failed to have prednisolone doses reduced below 7.5 mg per day, therefore not fulfilling the definition of remission. In all, 10 of 104 (9.6%) had continuous grumbling disease and did not reach a BVAS of 0.

A total of 35 major relapses occurred in 21 patients (14%; 11 heart, 10 respiratory tract, seven PNS, five gastrointestinal tract, two CNS). In all, 87 minor relapses in 42 patients (28%) were documented mainly affecting the ENT tract, the musculoskeletal system or appearing as constitutional symptoms. Major relapses tended to occur in the first 24 months after attaining first remission. Minor relapses occurred continuously during the whole period under evaluation.

Refractory disease was seen 19 times in 16 patients (10.6%; seven heart, four respiratory tract, three polyneuropathy (PNP), three gastrointestinal tract, two myositis).

The extent of the disease as measured by the DEI increased from onset of disease to referral to our centre at a mean rate of 6±0.9 DEI score per year. Since referral, the increase in DEI score decelerated significantly to a rate of 0.2±0.1 per year. This difference was highly significant (p<0.001; figure 1). The vasculitis damage index increased from median 1 (range 0–5) at initial presentation to median 2 (range 0–7) at end of follow-up. Table 2 shows cumulative manifestations as documented during the whole follow-up.

Table 2

Cumulative manifestations during follow-up

Treatment and side effects

During the follow-up time, a median of three immunosuppressants in addition to prednisolone was used (range 0–9). In all, 107 (71%) received CY (44 oral, 29 pulse, 34 both). The mean cumulative CY dose at end of follow-up was 28±3.4 g (range 1–210 g) including 57 who had been treated with CY before first referral (dose range 1–210 g). CY-induced cystitis was observed in six, one of whom developed bladder cancer after receiving 120 g CY. Six further malignancies (three prostate, one breast, one multiple myeloma, one colorectal) and four non-melanoma skin cancer were diagnosed.

The use of medium potent immunosuppressants was as follows: MTX 105, AZA 49 and LF 23. In refractory situations, the following drugs were used: interferon-α in 17, anti-TNF-α in 10 and rituximab in one patient. In five patients, IVIG was used in situations of concomitant infection and disease activity.

The estimated median cumulative prednisolone dose was 8.625 g (range 48.15–0.6 g) resulting in a mean daily dose of 7.8 mg. The prednisolone dose was within the targeted range (ie, ≤7.5 mg/day) during 69% of the observation time (figure 2). The median prednisolone dose at the end of follow-up was 5 mg per day (range 0–50) with a mean of 8.3 mg±9. In nine patients, the end of follow-up fell in a period of high disease activity due to relapse necessitating increased prednisolone doses. When excluding those from the analysis, a mean prednisolone dose at the end follow-up of 6 mg±2.7 resulted.

Figure 2

Kaplan–Meier plots of the survival of the whole cohort (A) and grouped according to eosinophilic granulomatosis and polyangiitis-related heart failure (B). SMR, standardised mortality ratios.

Overall, 31 severe infections, that is, requiring hospitalisation, were recognised, 10 of which occurred at initial presentation. Bacterial pneumonia was the most frequent infection (12 cases), followed by herpes zoster in five cases. Two CMV infections and one Pneumocystis jieroveci pneumonia were diagnosed. For details, see online supplementary table S5. GC dose was identified as a major risk factor: the mean prednisolone dose at diagnosis of the infection was 36±9 mg/day. A prednisolone dose above 10 mg/day was significantly associated with infectious complications (p<0.001). No other risk factor was identified.

In 58 (39%), osteoporosis was detected during follow-up of which 11 suffered a bone fracture. Other steroid-related side effects were cataract (21; 14%) and steroid-induced diabetes (12; 8%).


Using the additional data, the follow-up information was available in 142 patients and follow-up time increased to 92±5 months. Due to the uncertainty of these data, the only information used was whether or not a patient had died. Eight were lost to follow-up.

Twelve patients (four women) died 94±16 (mean±SD) months after diagnosis. Two died of acute worsening of pre-existing chronic cardiac failure (one with known heart involvement), two of sudden cardiac death (without prior evidence of cardiac involvement), one of myocardial infarction (pre-existing coronary heart disease classified as non-related to EGPA), two of non-controllable infections (pneumonia and septicaemia), one of asthma and one of colon carcinoma. In three, the cause of death remained uncertain. In the multivariate analysis, the only identifiable risk factor was heart failure (HR 3.37, p=0.049). An FFS ≥1 was associated with increased mortality (p=0.02) while ANCA status was not. There was a trend towards higher mortality in men (HR 2.93, p=0.12).

The standardised mortality ratio was 3.06 for patients with EGPA-associated heart failure and 1.29 for all EGPA, the latter not significantly different from the general population. The 5- and 10-year survival rates were 97% and 89%, respectively (figure 3).

Figure 3

The glucocorticoid dose target was 7.5 mg prednisolone or below per day. The graph shows months within the targeted range on the left side and out of this range on the right side for each patient. For 69% of the follow-up time the target was reached concerning the whole cohort. Few individual patients were out of range for a substantial part of their follow-ups.


In this report, we describe a better outcome of EGPA managed according to a predefined strategy when compared with published cohorts.2 ,3 ,16–20 Our algorithm, although established in 1990, is in accordance with the current recommendations given by the EULAR: we used CY plus GC in patients with an FFS >1 and MTX plus prednisolone in most patients without organ- or life-threatening EGPA. Although its efficacy has not formally been proven in EGPA, remission maintenance therapy was consequently applied and clear definitions of disease activity states were used.21

Our strategy consists of several details which have not been investigated in clinical trials: the repeated intensification of immunosuppression in cases of ongoing disease activity or need for high GC doses in non-organ- or life-threatening EGPA has not yet been evaluated. Furthermore, we used CY in patients with an FFS of 0 but with active PNS involvement, alveolar haemorrhage or severe eosinophilic alveolitis. This too, represents an aggressive treatment approach which can only be deduced from case reports or small series.22 ,23 This policy might bear the risk of increased adverse drug reactions. The same is true for the prolonged use of remission maintenance therapy, given that the appropriate duration of maintenance treatment is uncertain. The rationale for this was the belief that the most important risk factor for infections and other adverse outcomes is a high GC dose, either short-term or cumulative. This view is partially confirmed by the data presented here as well as by earlier observations in related diseases such as GPA.24 The overall rate of serious infections in our cohort was higher than that reported in the GPA cohort treated at our centre: 31 infections (29.8%) in 460 patient-years in EGPA versus 33 (19.8%) in 651 patient-years in GPA.8 The likely reason for this difference was that despite the same aggressive treatment approach remission was less likely to occur in EGPA (67% vs 74% in GPA) and more importantly relapses requiring increased GC doses were much more frequent in EGPA (30% major relapses and 60% minor relapses vs 35% total relapses). GC-dependence is one of the most urgent management problems. By the use of more intense immunosuppression, we achieved an acceptable low GC dose of median 5 mg/day at the end of follow-up. Compared with other cohorts with less strict steroid reduction and no use of maintenance therapy our results show a favourable outcome: In their study of 6 versus 12 CY pulses with 48 patients and a mean follow-up of 43 months, Cohen et al2 report major relapses of 56% and 62% in the two treatment groups. The rate of infection was 44% and osteoporotic bone fractures occurred in 17% (vs 10% in our cohort). The higher rates of remission reported in this and other earlier trials (91% and 84% for the two treatment groups) are probably due to the different definitions applied: besides the absence of any disease activity our definition (as the recently consented definition by the EULAR9) also required a prednisolone dose below 7.5 mg/day. An absence of disease activity was achieved in 90% in our study. Ribi et al reported that 42% failed to respond to steroid monotherapy or relapsed during steroid tapering. In that study an absence of disease activity was achieved in 79% at the end of follow-up but a substantial number of patients still used medium to high steroid doses.3 According to our experience, the use of additional immunosuppression seems reasonable even in most patients with an FFS of 0.

The structured patient education programme applied to each patient may also have contributed to improved outcomes. As shown earlier, such programmes increase knowledge and awareness and thereby might lead to earlier recognition of disease flares and drug adverse events.15

Our data show that several manifestations may remain unrecognised if specialists are not routinely involved. For example, we found the highest rate of ENT involvement (93%) yet published, whereas consultation of an ENT specialist only if specific symptoms were present led to much lower rates, for example, 53% in an actual French cohort.25 In that cohort, ENT involvement was associated with improved survival, a result not reproducible in our patients. Our data confirm a high, nearly 50%, rate of cardiac involvement, a number also described in a recently published systematic investigation using cardiac MRI among other methods.26 ,27 Other historic cohorts may have under-recognised cardiac involvement reporting frequencies of 0%,3 16%18 or 30%.17 In most published series, there is no information on eye involvement, which was seen in 12% here, again the highest frequency reported until now. Whereas intensification of immunosuppression for moderate ENT activity might still be a point of controversy, it seems likely that early detection of heart involvement and its consequent therapy can improve outcome. In accordance with an actual re-evaluation of the FFS,25 we found cardiac insufficiency to be the only independent parameter associated with increased mortality. We conclude that a structured interdisciplinary work-up involving vasculitis-trained specialists may contribute to a better outcome. For a detailed comparison of our findings with published other cohorts, see online supplementary table S6.

Obviously, there are several limitations to be considered when interpreting the data presented here: there is no control group and therefore the outcome of our management strategy can only be compared with that from other cohorts. Given a general medical progress, the latest cohort is likely to show better outcomes. However, the survival rates described here are even better than those for patients with an FFS of 0 as published very recently.25

Referral bias is another important point to consider: it might be argued that severely diseased patients could not be referred to our centre. On the other hand, it is plausible that more patients with difficult to treat EGPA may be referred to an expert centre. This question must been left unanswered as long as no registry systematically captures all data.

When analysing the management strategy applied, another difficulty is that changes in several procedures have been made over the years. For example, there was a shift from using primarily oral CY towards the use of pulse therapy in later years. Moreover, new drugs were included for salvage therapy as they became available.28 ,29 In addition, diagnostic progress has to be taken into account; for example, with the use of cardiac MRI it is likely that the sensitivity for detecting heart involvement increased.

Taken together, we describe a complex strategy for the management of EGPA with the major aspects being a structured interdisciplinary and repeated work-up at an experienced centre, a treatment algorithm including regular adaption of medication if the predefined target of remission which included a GC dose ≤7.5 mg/day was not achieved and a structured patient education programme. This approach was effective leading to normalisation of life-expectancy in most of the patients. Treatment at an experienced vasculitis centre with a defined management strategy as presented here should therefore be the standard of care in EGPA.


View Abstract

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    • Supplementary Data

      This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

      Files in this Data Supplement:


    • FM and JPB contributed equally.

    • Contributors FM: conception, data acquisition and analysis, writing. JPB: data acquisition and analysis. BH, CMe and WLG: conception, writing. KH-U: pathology, data acquisition. ML: ENT, data acquisition. CMa: statistics. BN: ophthalmologist, data acquisition. GR: cardiologist, data acquisition.

    • Competing interest None.

    • Provenance and peer review Not commissioned; externally peer reviewed.

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.